Elgie What Is Semi-Presidentialism
Elgie What Is Semi-Presidentialism
Elgie What Is Semi-Presidentialism
where is it found?
Robert Elgie
In the years that followed, most people adopted the basics of Duverger’s
understanding of semi-presidentialism. However, some writers altered the
wording of his definition and/or the application of the concept. For example,
Sartori (1997, 131–2) argued that a system was semi-presidential if it had the
following five characteristics:
the Austrian and Icelandic presidents (not the Irish one) are strong only on
paper, that is, are constitutionally given powers that the living constitution
relegates to inanation. . . . But when the material constitution deprives a
president of powers that remain a dead letter of the formal constitution, then
a ‘dead element’ surely cannot establish the nature of a political form and
the class to which is belongs.
(Ibid.)
In other words, for Sartori the powers of the president are integral to the under-
standing of the term. Moreover, what matters is not the powers that presidents
enjoy on paper, but the powers they wield in practice. In order to determine
whether or not a country is semi-presidential, he argues, we need to know how
power is actually exercised in a country. If we find that there is a basic balance
of power between a directly elected president and a prime minister responsible
to parliament, then we can class the country as semi-presidential. If not, then we
should class the country as either parliamentary if the president is weak or presi-
dential if the president is strong.
In other work I have argued that this way of defining semi-presidentialism is
problematic. In particular, I have argued that any mention of the respective
powers of the president and prime minister should be eliminated from the defini-
tion of the term (Elgie 1999, 2005). The reason why is that the inclusion of any
mention of powers allows different writers to come up with different lists of
semi-presidential countries. For example, writing before the wave of democrat-
ization in the early 1990s, Lijphart (1992, 9) implied that France was the only
semi-presidential democracy in the world. Writing around the same time, Stepan
and Skach (1993, 3), who follow Duverger’s definition, identified two semi-
presidential countries, France and Portugal. By contrast, writing in the same
period, Sartori argued that France and Finland were semi-presidential, that Por-
tugal had been semi-presidential from 1976 to 1982, and that Sri Lanka was at
the edge of presidentialism and semi-presidentialism (Sartori 1997, ch. 7).1
When we go beyond this early work the disagreement as to which countries
should be classed as semi-presidential only increases. This can be seen very
clearly when writers started to examine the process of democratization in
Central and Eastern Europe and the former Soviet Union. For example, Easter
(1997, 190) identified only three mixed systems in all the countries in this region
– Lithuania, Moldova and Poland – plus Mongolia. By contrast, focusing only
on Central and Eastern Europe and even then excluding a number of potential
candidates such as Macedonia and Moldova, von Beyme (2001, 16) identified
no fewer than eight semi-presidential countries – Belarus, Croatia, Lithuania,
Poland, Romania, Russia, Serbia2 and Ukraine. Indeed, Bahro et al. (1998, 208)
went further still, stating that ‘the majority of post-communist states on the terri-
tory of the former Soviet Union plus Mongolia, as well as Poland, Romania,
What is semi-presidentialism? 5
Bulgaria and also Slovenia, Croatia, Serbia and Macedonia comply with this
scheme’.
The list of semi-presidential countries varies so much from one writer to
another because each writer has free rein to decide what constitutes the level
of ‘quite considerable [presidential] powers’ that are required for semi-
presidentialism. As a result, some writers interpret semi-presidentialism in a way
that excludes countries such as Austria, Bulgaria, Iceland, Ireland and Slovenia,
where the president’s powers are deemed to be not considerable enough. This
would be consistent with Sartori’s position. Other writers interpret semi-
presidentialism in a way that excludes those countries as well as other countries,
such as Belarus and Russia, where the president’s powers are considered to be
too considerable. This would be consistent with the position of Stepan and
Skach. Yet other writers are happy to include all these countries, presumably on
the assumption that the term ‘quite considerable’ covers a vast range of possible
powers. This is the approach taken by Bahro et al. If we gathered together every
list of semi-presidential countries in existence, then we might be able to identify
a core of apparently ‘uncontroversial’ examples – probably France, Mongolia
and Poland and perhaps Lithuania too. However, even if we were able to
identify a common set of countries, we would not have a common definition on
which to base the list. In sum, including a clause along the lines that presidents
must enjoy ‘quite considerable powers’ in the definition of semi-presidentialism
means that there is an inherent subjectivity in the way people identify examples
of the concept.
The fact that there is a basic subjectivity in the way many people define semi-
presidentialism is of more than just semantic importance. This is because it
affects the judgements people make about the respective advantages and
disadvantages of this regime type and, indeed, others too. For example, if we
define semi-presidentialism so as to include the majority of countries in the
former Soviet Union as semi-presidential, including countries such as Azerbai-
jan, Belarus, Kazakhstan, Russia, Uzbekistan and so on, then the record of this
type of government looks poor. However, if we define semi-presidentialism so
as to exclude these countries as semi-presidential and include only ‘uncontrover-
sial’ countries such as Lithuania, Poland and Romania, then semi-presidentialism’s
record looks much better. Even so, if we do exclude them, then naturally enough
the record of presidentialism looks worse because countries such as Azerbaijan,
Belarus, Kazakhstan, Russia and Uzbekistan are invariably classed as presiden-
tial if they are not classed as semi-presidential. In other words, the fact that the
list of semi-presidential countries varies so much from one writer to another
means that often like is not being compared with like. This situation compro-
mises the general debate about institututional design. We do not have a suffi-
ciently large n for this sort of definitional problem to wash out in any statistical
analysis. Thus, how can we say that presidentialism is worse than semi-
presidentialism, or vice versa, or that they are equally bad and that parliamen-
tarism is better, when each writer has a different understanding of these concepts
and bases his/her conclusions on a different set of examples?
6 R. Elgie
Worse still, by including the criterion of ‘quite considerable powers’, or an
equivalent clause, in the definition of semi-presidentialism, writers may unwit-
tingly predetermine their judgements about the concept. For example, a seem-
ingly ‘commonsense’ way of understanding semi-presidentialism is to say that it
is a mixed system where there is a basic balance of presidential and prime minis-
terial powers. I have argued above that this is the sort of logic Sartori adopts.
However, if we define semi-presidentialism in this way, then we should not be
surprised to find that semi-presidential countries are often associated with intra-
executive conflict. In consolidated democracies, such conflict may lead to polit-
ical deadlock. In fragile democracies, it may lead to either the president or the
prime minister, or both, trying to go beyond the constitution and perhaps seeking
the support of the military in order to assert control. Either way, the result may
be either inefficient and/or destabilizing decision making and may lead writers
to recommend that semi-presidentialism should be avoided. The problem,
though, is that this judgement is based on a logic that comes very close to being
circular. If we define semi-presidentialism as the situation where there is an
actual balance of power between the leading political figures in the system, then
we should not be surprised to find that there is in fact a balance of power
between them as well as all that goes with it. So, by including reference to
powers in the definition of semi-presidentialism, we may end up with a major
problem of endogeneity. In other words, we may be putting the methodological
cart before the horse.
In the introduction to this chapter I made the point that the debate about
regime types has important ‘real world’ implications. Given these implications,
we need to try to make sure that this debate is conducted as rigorously as pos-
sible. I argue that we can do so only if we exclude all reference to the powers of
presidents and prime ministers from any definition of semi-presidentialism.3
Therefore, I have proposed the following definition:
On the basis of this definition, we only need to read the constitution of a country
in order to determine whether or not it is semi-presidential. We do not need to
know how powers are exercised in practice. Moreover, we only need to read
very limited aspects of a constitution in order to determine whether or not a
country is semi-presidential. Is the president directly elected and is the govern-
ment responsible to parliament? If so, then the country is semi-presidential. If
not, then it belongs to another category. All of this means that it is much easier
to establish an ‘uncontroversial’ list of semi-presidential countries than if we
include reference to any amount of presidential and prime ministerial powers as
a criterion. Moreover, by defining semi-presidentialism in this way, we do not
prejudge the way in which semi-presidential countries operate. As we shall see,
the list of semi-presidential systems includes countries where the president has
What is semi-presidentialism? 7
great powers, countries where the president has few powers and countries where
there is a balance of presidential and prime ministerial powers. Thus, we avoid
the potential for endogeneity problems associated with case selection. Overall,
we arrive at a much less subjective way of defining semi-presidentialism and
one which gives us greater potential to make rigorous judgements about the
advantages and disadvantages of this type of regime.
That said, we cannot eliminate all subjectivity from the definition of semi-
presidentialism. The nature of constitutional law is such that some countries do
not fall neatly into one category or another. For example, in Slovakia the presid-
ent can be recalled by a popular referendum. So, strictly speaking, the president
does not necessarily serve for a fixed term. In this case, we have to decide
whether we continue to classify Slovakia as semi-presidential or whether we
create a new category for Slovakia and countries like it. My judgement call is
that we should continue to classify Slovakia as semi-presidential. Unlike the
direct election of the president and the responsibility of the government to the
legislature, a recall vote is intended to be an extraordinary and, hence, arguably
non-constitutive political feature. True, another writer may argue that it is con-
stitutive and choose to create a new category for Slovakia and other countries
that have this unusual constitutional feature. However, even if s/he did so, at
least the guidelines would still be clear as to when a country should be classed
as semi-presidential or when it should be placed in the new category alongside
Slovakia. Indeed, if we made this choice, then we would place Iceland alongside
Slovakia in this separate category. Thus, the classification process would still be
reasonably objective. The same is not true if we include powers as part of the
definition of semi-presidentialism. In that case, we have no guidelines as to
where the thresholds between parliamentarism, semi-presidentialism and presi-
dentialism lie. As a result, individual writers are able to make their own choices
and very varied lists of countries appear. So deciding to classify Slovakia as
semi-presidential even though its president can be subject to a recall vote is an
inherently less subjective exercise than deciding to classify Slovakia as
parliamentary, presidential or semi-presidential as a function of how many
powers the president has. Even though there is still ambiguity as to how we
should classify Slovakia on the basis of my understanding of the term, the ambi-
guity is of a different nature from definitions that classify countries with refer-
ence to the actual powers of political actors.
There are other judgement calls to be made when adopting my definition of
semi-presidentialism. For example, in Argentina the head of the government is
responsible to the legislature, but the cabinet is not. Again, a judgement call
must be made. In this case, I exclude Argentina from the list of semi-presidential
systems because there is no collective cabinet responsibility. Others, though,
may wish to ignore the individual responsibility of the head of government and
classify Argentina as presidential or create a new category for countries where
there is individual but not collective responsibility. A further case involves a
number of countries such as Azerbaijan, Mozambique and South Korea. In these
countries the constitution states that the prime minister is appointed with the
8 R. Elgie
consent of parliament, but that parliament may only recommend the removal of
the prime minister and the president may decide to ignore the parliamentary
vote. Thus, it appears as if the prime minister and cabinet are not really respons-
ible to the legislature at all. However, here I include these countries as
semi-presidential because the need for parliamentary consent implies that
constitutionally such consent may be withheld. Thus, these countries are very
close to the standard requirement for semi-presidentialism whereby the legis-
lature has some say in determining the life of the government. It may well be the
case that in some or all of these countries and others like them parliament may
be the puppet of the president and may in fact never have even contemplated
withholding its consent to a presidential nominee. Even so, in order to be as
consistent as possible, I do not classify countries by way of reference to how
powers are exercised in practice. As a result, I classify these countries as semi-
presidential.
On the basis of this definition, I calculate that there are currently fifty-five
semi-presidential countries in the world. (See Table 1.1 for the list of semi-presi-
dential countries as of 1 January 2006.) Clearly, in some of these countries the
democratic system is purely nominal. So I am not implying that there are fifty-
four democratic semi-presidential countries. I am simply saying that according
to available constitutional sources, and having made a number of constitutional
judgement calls, all of these countries currently have at least a basic form of
semi-presidentialism. Other countries have been semi-presidential at one time
but have since chosen a different form of government. The most famous
example is the Weimar Republic in inter-war Germany. More recently, Moldova
was semi-presidential from 1991 to 2000, as were the Comoros from 1978 to
1984 and 1992 to 1999. Other countries have fluctuated between different
regime types. For example, Senegal was presidential from 1963 to 1970, semi-
presidential from 1970 to 1983, presidential from 1983 to 1991 and semi-
presidential since that time. Other countries may adopt semi-presidentialism at
some future point. In December 2005 voters in the Democratic Republic of
Congo approved a semi-presidential constitution that took force in late 2006.
The draft constitution for a state of Palestine drawn up under the auspices of the
Palestinian National Authority in 2003 makes provision for a semi-presidential
form of government. Whatever may or may not happen in the future, around 30
per cent of all independent states currently have a semi-presidential form of
government.
There is a fundamental objection to my approach, namely that the list of
countries classed as semi-presidential on the basis of my definition is counter-
intuitive. There are three versions of this objection. The first version comes from
country specialists. These are the people who are sometimes transformed in dis-
cussion from being polite and unassuming political scientists into area studies
specialists who are red with rage. For specialists on Irish politics, Ireland is
parliamentary. It simply cannot be semi-presidential. It goes against everything
that for decades everyone has been trained to understand about Irish politics. By
the same token, for most Russian experts, Russia is not semi-presidential. It is
What is semi-presidentialism? 9
Table 1.1 Semi-presidential countries
Notes
1 Sartori also states that Russia became semi-presidential in 1993, but that, like Sri
Lanka, it stands at the ‘extreme edge of the category’ (1997, 138 n. 9).
2 The 1990 constitution of the Republic of Serbia is semi-presidential. However, until
2003 the Republic of Serbia was part of the Federal Republic of Yugoslavia and since
2003 it has been a constituent part of the State Union of Serbia and Montenegro.
Neither the Federal Republic of Yugoslavia nor the State Union of Serbia and Mon-
tenegro is semi-presidential.
3 Indeed, I argue that the same point applies to the definition of all regime types, includ-
ing presidentialism and parliamentarism (Elgie 1998).
4 If my interpretation of Duverger is correct, then it implies that he ignored the clause in
his definition about presidents needing ‘quite considerable powers’, or that he inter-
preted the clause very loosely.